FALLBACK in Teradata

An active data warehouse with fallback enabled provides recovery from pathological failure scenarios.

The ability of the Teradata RDBMS to maintain an extra copy of every row of a table in different AMPs. During normal processing, reads are performed using only the primary copy, while updates are performed on both copies. If a failure renders the primary copy inaccessible, the fallback copy is used for reads and updates. The database maintains a log of changes missed by inaccessible copies and automatically applies the missed changes to synchronize the table copy when it becomes accessible.

FALLBACK tables use twice as much disk space as NON-FALLBACK rows.
Generally, fallback is usually only used on critical tables since it will double your disk usage.

During table creation or after a table is created, we may specify whether or not the system should keep a Fallback copy of the table.

CREATE MULTISET TABLE employee, fallback,
datablocksize=32 kbytes,freespace= 10 percent,checksum=none
(empno integer,
salary integer); 

CREATE SET TABLE deptartment,No fallback,
No Before Journal, No After  Journal,
Checksum =default
employee_number integer,
deptno integer